本文へ移動

AIモデルを比較

数値表
モデル・推論設定解決率費用/試行(USD)時間(分)
GPT-6 Astra (max)58.29.900546.6
Fable 5.1 (max)57.918.919764.9
GPT-6 Astra (xhigh)57.97.122836.8
GPT-6 Astra (high)57.96.877035.2
GPT-6 Astra (medium)54.25.802431.4
Opus 5 (max)51.818.088279.9
GPT-6 Astra (low)50.64.719128.1
Fable 5 (max)44.622.015270.0
GLM-5.3 (max)41.88.265597.2
GPT-5.6 Sol (max)37.37.702139.8
Opus 4.8 (max)23.619.640286.1
GPT-5.6 Terra (max)21.55.253141.9
Grok 4.6 (high)20.310.883639.2
Gemini 3.8 Flash (high)19.15.541733.3
GPT-5.6 Luna (max)17.31.050568.1
Grok 4.5 (high)12.46.345852.7
Sonnet 5 (max)12.429.1026108.5
Gemini 3.7 Flash (high)11.23.823828.1
Fable 5.1 (xhigh)57.914.763853.4
Fable 5.1 (high)54.612.076950.3
Fable 5.1 (medium)53.98.584542.1
Fable 5.1 (low)43.37.147638.2
Claude Sonnet 5.5 (low)20.00.75724.6
Claude Sonnet 5.5 (medium)28.80.83474.9
Claude Sonnet 5.5 (high)43.01.94018.0
Claude Sonnet 5.5 (xhigh)61.55.298017.2
Claude Sonnet 5.5 (max)70.612.540628.2
GPT-6 Sol (low)9.10.49003.3
GPT-6 Sol (medium)18.71.12005.6
GPT-6 Sol (high)26.31.60007.8
GPT-6 Sol (xhigh)30.31.910010.3
GPT-6 Sol (max)43.94.000018.0
Claude Opus 5.5 (low・フォールバックあり)31.32.08005.0
Claude Opus 5.5 (medium・フォールバックあり)52.54.04009.9
Claude Opus 5.5 (high・フォールバックあり)56.65.120012.6
Claude Opus 5.5 (xhigh・フォールバックあり)59.68.780020.1
Claude Opus 5.5 (max・フォールバックあり)59.613.110027.5
Gemini 3.8 Flash (low)10.15.38009.2
Gemini 3.8 Flash (medium)19.74.61008.6
GPT-6 Luna (low)0.00.01000.4
GPT-6 Luna (medium)2.50.07004.2
GPT-6 Luna (high)4.50.12008.0
GPT-6 Luna (xhigh)8.10.200011.0
GPT-6 Luna (max)12.60.240014.1
比較するモデル